That is why the Lancet study used statistical sampling techniques to estimate the actual numbers. These techniques have been widely used in other similar situations, and the Lancet results indicate that the actual figures are now near 1,000,000.
The second survey<2><3><4> published on 11 October 2006, estimated 654,965 excess deaths related to the war, or 2.5% of the population, through the end of June 2006. The new study applied similar methods and involved surveys between May 20 and July 10, 2006.<4> More households were surveyed, allowing for a 95% confidence interval of 392,979 to 942,636 excess Iraqi deaths. 601,027 deaths (range of 426,369 to 793,663 using a 95% confidence interval) were due to violence. 31% of those were attributed to the Coalition, 24% to others, 46% unknown. The causes of violent deaths were gunshot (56%), car bomb (13%), other explosion/ordnance (14%), air strike (13%), accident (2%), unknown (2%).
The Lancet surveys are controversial largely because their mortality figures are much higher than those in other reports that used different methodologies, including those of the Iraq Body Count project, the United Nations, and the Iraqi Ministry of Health. Out of all the Iraqi casualty estimates so far,
the Lancet survey is the only peer-reviewed one. The Lancet surveys have been supported by many epidemiologists<5> and statisticians, as well as the September 2007 ORB survey. ORB used a survey method, as did the Lancet authors, and ORB also produced a high estimate of Iraqi deaths. However, the Lancet surveys have been criticized by the US and Iraqi governments, the Iraq Body Count project, epidemiologists, demographers, Iraq-war journalists and others. See the sections below on "criticisms".
http://en.wikipedia.org/wiki/Lancet_surveys_of_mortality_before_and_after_the_2003_invasion_of_Iraq